Correlations between family history and cancer characteristics in 2256 breast cancer patients

A comparison of 692 early invasive breast cancer with, and 1564 without, a family history of breast cancer showed that the former were younger at diagnosis (P=0.002), had smaller tumours (P=0.012), were more frequently oestrogen receptor positive (P=0.006) and diagnosed preclinically (P<0.001).


MATERIALS AND METHODS
A total of 2256 women with early invasive operable breast cancer, who underwent surgery at Verona Hospitals between January 1992 and April 2001, were asked at their first visit whether they had any first-or second-degree relatives who had had breast cancer. Our analysis first compared those reporting at least one affected relative (FH þ ) with those reporting no affected relative (FHÀ); subsequently, the FH þ patients were divided into those with at least one first-degree relative (1st DFH) and those with only second-degree relatives with breast cancer (2nd DFH): only firstand second-degree relatives were considered in order to reduce ascertainment bias. Answers were always checked at the subsequent visit (at the time of the first cycle for the patients receiving chemotherapy, after 3 months for the others).
Immunohistochemistry (IHC) defined oestrogen (ER) and progesterone receptor (PgR) status and was considered positive if more than 10% of the cells were stained for either. The replicative cell fraction was IHC stained using the Ki-67 monoclonal antibody (Mab-DAKO-PC); given the lack of an accepted cutoff point, the results were arbitrarily classified as low, medium or high (p10%, 11 -25% or 425% of stained cells). C-erbB-2 levels were determined by IHC using the DAKO-PC monoclonal antibody, and considered positive if at least one cell was stained.
At their first visit to the Department of Medical Oncology, all patients were asked about their disease presentation and divided into those who underwent mammography and ultrasonography because of breast discomfort or a lump (symptomatic) and those without any subjective symptoms (asymptomatic). A public screening programme has been active in Verona since July 1999, before which 30% of our patients had undergone mammography and ultrasonography even in the absence of subjective symptoms.
The w 2 test was used to compare the prevalence of FH þ (1st and 2nd DFH) and FHÀ women in relation to all the variables.
As it is known that younger women have a higher risk associated with a family history, an age-adjusted analysis was made to compare the prevalence of FH þ women across the categories of the other considered variables.
Significance was tested using the likelihood ratio statistic and a significance level of 0.05.

RESULTS
The overall prevalence of FH þ women was 30.7% (692), of whom, 356 (51.4%) were classified as 1st DFH and 336 (48.6%) as 2nd DFH. Table 1 shows the crude and age-adjusted prevalence of FH þ women across the various categories. Family history was significantly associated with tumour size, ER, age at diagnosis and disease presentation. No significant association was found between family history and nodal involvement, PgR, Ki-67 status, grading or c-erb B2 levels (Table 1).
Associations with degree of familial relationship were investigated only for the variables significantly associated with a family history when a more general definition was used (tumour size, ER, age at diagnosis and disease presentation). Only age at diagnosis was significant (w 2 ¼ 36.0; Po0.001): older women were more likely to have a first-degree relative (76.1% of those aged 474 years had an affected first-degree relative vs 51.9% of those aged 45 -74 and 33.0% of those aged o45 years; data not shown).

DISCUSSION
A positive family history is now an established risk factor for breast cancer, but few studies have focused on the characteristics of breast cancer in FH þ women. We have done so because physicians can rarely study genetic patterns in clinical practice, and must therefore rely on patient-supplied FH reports despite their possible inaccuracy (Slattery and Kerber, 1993;Eerola et al, 2000). Furthermore, some tumours will develop in relatives after the patient has been examined.
We have only considered first-and second-degree relatives, the percentages of which were similar to those reported in other studies (Collaborative Group on Hormonal Factors in Breast Cancer, 2001). Our adjusted results suggest significant differences in FH þ tumours, which seem to be smaller, more often ER þ , and more likely to be diagnosed at a younger age and at an asymptomatic stage. The published data are few and inconsistent, and usually not subject to multivariate analysis adjusted for age; in fact, only Kreiger about receptor status (Kreiger et al, 1991) and Swede about c-erbB2 (Swede et al, 2001) compared FH with clinical and pathological characteristics in a multivariate analysis adjusted for age as in this paper. Most have not found any significant difference in tumour size between FH þ and FHÀ patients (Fukutomi et al, 1993;Israeli et al, 1994;Tsuchiya et al, 1998;Russo et al, 2002), whereas Mohammed observed a trend (Mohammed et al, 1998) and Colditz reported a higher percentage of T1 in women FH þ (60%) vs FHÀ (54%) (Colditz et al, 1993). There were no significant differences in nodal involvement between our FHÀ and FH þ tumours. Some studies (Fukutomi et al, 1993;Mohammed et al, 1998) reported that FH þ patients were more likely to have tumours with fewer nodal metastasis, but others (Ruder et al, 1988;Israeli et al, 1994;Tsuchiya et al, 1998;Russo et al, 2002) found no significant difference. Our univariate and multivariate analyses showed that FH þ tumours are more likely to be ER þ , but there were no statistically significant differences in PgR, Ki-67 or grading. Some studies found that FH þ tumours are more likely to be ERÀ and PgRÀ (Kreiger et al, 1991;Huang et al, 2000), have a higher proliferation rate (Marcus et al, 1994) and higher grading (Mohammed et al, 1998), but none of the others found any differences (Fukutomi et al, 1993;Israeli et al, 1994;Tsuchiya et al, 1998;Russo et al, 2002).
In agreement with others (Fukutomi et al, 1993;Swede et al, 2001), we did not find any differences in relation to the more recent c-erbB2 marker.
Our FH þ tumours were more frequently diagnosed in an asymptomatic phase than the FHÀ tumours. This is probably due to the greater sensitisation induced by having one or more affected relatives, which encourages asymptomatic women to undergo diagnostic investigations; our data are similar to those of others (Colditz et al, 1993;Murabito et al, 2001) who have observed that FH þ women undergo more mammographies.
Our FH þ patients were younger at diagnosis, which confirms some previous data (Israeli et al, 1994), but not others (McCredie et al, 1997;Tsuchiya et al, 1998;Russo et al, 2002).
In conclusion, FH þ seems to be associated with an asymptomatic diagnosis (and therefore smaller and ER þ tumours), possibly because of the more intensive use of mammography or differences in biological behaviour. FH þ patients are younger, although it is not clear as to whether this reflects a truly earlier disease onset or an earlier diagnosis due to the more intensive use of mammography.